Connors ETF strategy 1b: Three day high/low – short entry

by Chris White on February 7, 2010

————————————————————————————————–
Article Update: Find these ETF Signals with my new program
————————————————————————————————–
In the previous EdgeRater strategy analysis article I analyzed the first of Larry Connors ETF strategies – the Three day high/low method using a long entry. That strategy appeared to test well and gave statistically advantageous entries when tested over the historical period.

In this article I will analyze the short entry method of the same strategy. If the short entry method has a positive expectation in addition to the long entry method then there is something of value here.

So, let’s get started.

Symbol List

I will be using the same set of symbols as in the previous article. To use this list within EdgeRater, create a text file containing the symbols and names (copy and paste the list in the previous article to a new text file) and add it to the {EdgeRater}\Symbols Lists directory in your documents folder.

Historical testing period

The testing period for Larry’s strategies is 1/1/1993 to 12/31/2008.

EdgeRater snapshot

For this analysis I have used the Yahoo! data provider and created my data snapshot using the above symbol list and date range. I was able to just re-use the snapshot that I created for the first article because the only differences between the two strategies are the entry rules.

Entry Rules

Larry’s entry rules for the short trade are the reverse of the long trade entries and are as follows:

  • Today’s close price is below the 200 day SMA
  • Today’s close price is above the 5 day SMA
  • The high price of 2 days ago is above the high price of 3 days ago
  • The low price of 2 days ago is above the low price of 3 days ago
  • The high price of 1 day ago is above the high price of 2 days ago
  • The low price of 1 day ago is above the low price of 2 days ago
  • The high price of today is above the high price of 1 day ago
  • The low price of today is above the low price of 1 day ago

EdgeRater Chart Script

The entry rules can be written in Chart Script like so:


Below200:= C < SMA(C,200);
Above5:= C > SMA(C, 5);
HL3:= (Ref(H, 2) > Ref(H, 3)) & (Ref(L, 2) > Ref(L, 3));
HL2:= (Ref(H, 1) > Ref(H, 2)) & (Ref(L, 1) > Ref(L, 2));
HL1:= (H > Ref(H, 1)) & (L > Ref(L, 1));

Event: Below200 & Above5 & HL3 & HL2 & HL1;

Running the Analysis

Once the Chart Script has been entered into EdgeRater it will appear as a Chart Script selector in the Edge Runner tab (because it has an output line named ‘Event’). To run this analysis over the prepared snapshot of data, first ensure that the snapshot has been chosen in the Snapshot pane and then make sure to select the ‘Use global exclusion lists’ check box in the Snapshot Filters. This will ensure that the _INDEX symbol that is automatically added to every snapshot will not be considered during the analysis run.

Once everything has been set up, hit the ‘Run’ button to start the analysis.

————————————————————————————————–
Article Update: Find these ETF Signals with my new program
————————————————————————————————–
Results

So, what do these setups look like? Here are a couple of examples – one that fulfilled the premise and one that didn’t:

GLD ETF successful entry

GLD ETF successful entry

Successful entry point in the GLD ETF – following bars show a sell-off

GLD ETF unsuccessful entry

GLD ETF unsuccessful entry

Unsuccessful entry point in GLD ETF – following bars show a rally

In both cases you can see that the setup conditions were met – three consecutive days of higher highs and higher lows with stock closing on the signal day below the 200 day SMA and above the 5 day SMA. Clearly not every trade will be a winner, but it is important to know the statistics of this strategy to see if it is a strategy that is a) profitable in the long run, and b) fits your trading personality so that you are comfortable trading it.

Statistics

The Next N Bars report shows that this short entry was profitable during the analysis period when holding anywhere from 1 through to 10 days (I didn’t examine any longer than a 10 day hold) with a 5 day hold period performing best at an average -0.71% loss. In order to compare to the long entry strategy of the previous article I will show the histogram of the 4 day hold:

4 day hold statistics

4 day hold statistics

3 day High/Low short entry with 4 day hold period

Baseline

The baseline scan is identical to the one in the previous article due to identical symbol list and timeframe, I have repeated the result here for convenience.

Next N Bars baseline

Next N Bars baseline

Baseline scan mean gain of +0.0944% with standard deviation of 3.4776

Summary
————————————————————————————————–
Article Update: Find these ETF Signals with my new program
————————————————————————————————–
Larry Connors’ three day high/low strategy – short entry method had a positive expectation over the testing period and interestingly was profitable in addition to the long entry method using the same time period and set of ETFs.

If this strategy is something that fits your personality you should definitely enter it into EdgeRater and experiment with the results for yourself. Once you have seen the various instances of setups and looked at all of the reports I have a couple of suggestions for further analysis.

  • In the setup instance charts above you can clearly see that even though all the entry conditions were met in the failure case, the 200 day SMA was in fact moving higher indicating that the long term trend could be changing. An extra entry condition worth looking at would be that the 200 day SMA has to have a downward slope.
  • In practice it would not be possible to enter positions at the close of the day when the conditions required to determine the entry rely on the actual closing price. The Next N Bars report in EdgeRater can be set to a baseline price of the following day’s open. An extra analysis step could be to experiment with these different baseline prices by changing the radio button setting.

Once you have a strategy that you want to trade, run EdgeRater each evening to update the snapshot with the latest data and run the selector over that snapshot. The last column in the event grid will contain the symbols that triggered for trading tomorrow.
This analysis is presented as information only and is not a recommendation to trade any particular symbol or strategy. All trading involves risk and you alone are responsible for your trades.

{ 2 comments }

Sample code – Hammer and EMA cross

by Chris White on February 3, 2010

An EdgeRater user recently asked about using the candlestick pattern library and combining one of the candlestick patterns with another condition such as the close price crossing over the EMA.

This sample code shows one way to achieve that using the Managed Code API.

To add this code to your EdgeRater insallation just create a file in the {EdgeRater}\Selectors\Series directory called Hammer.cs and copy and paste the code into that file. The next time you refresh the ‘Managed Code’ library, Hammer.cs will be discovered and you can drag and drop it onto the selector design surface. It will need to be compiled before it can be run.

using System;
using System.ComponentModel;
using EdgeRater.Edge;
using EdgeRater.Edge.TechnicalAnalysis;

namespace EdgeRater.UserCode
{
    [Serializable]
    [Description("Hammer and Close cross over EMA")]
    [DisplayName("Hammer EMA cross")]
    public class Hammer : SeriesSelector
    {

        //
        // This method is called for each bar of each series in the
        // data snapshot. If you return true, an event will be created
        // for that bar and will be shown in the event grid
        //
        public override bool Execute(BarSeries series, SelectorContext context)
        {
            DataSeries hammer = CandleFunctions.Hammer(series.Open, series.High, series.Low, series.Close);

            DataSeries ema = Functions.Ema(series.Close, 22);

            bool isCross = SelectorHelper.CheckCrossCondition(series.Close, CrossCondition.CrossesFromBelow, ema);

            bool isHammer = !double.IsNaN(hammer[0]) && hammer[0] != 0;

            return isCross && isHammer;
        }

        //
        // The value of this property should be the number of prior
        // bars needed to perform the selection. For instance, if in
        // the Execute() method you have used a function that needs
        // 5 bars before it can calculate a result, and another that
        // needs 10 bars, you should return a value of 10 for this
        // property.
        //
        public override int Lookback
        {
            get
            {
                return (Math.Max(CandleFunctions.HammerLookback(), 22));
            }
        }
    }
}

{ 1 comment }

Connors ETF strategy 1: Three day high/low method

by Chris White on January 20, 2010

————————————————————————————————–
Article Update: Find these ETF Signals with my new program
————————————————————————————————–

This is the first in a series of articles which will analyze the strategies contained in Larry Connors’ popular 2009 book Larry Connors’ High Probability ETF Trading which presents 7 strategies for trading highly liquid ETFs. The goal is to determine the statistical significance of the entry points. In other words I want to find out if the entries presented in the book were any better than random over the time period tested.

I will be using EdgeRater to analyze the strategies and this article will present the steps I use along the way so that anyone can follow along and maybe take the analysis further.

3-Day High/Low Method

The first strategy in the book is the 3-Day high/low method. This strategy looks for ETFs that have made lower highs and lower lows for three consecutive days. Combined with a long term moving average to ensure the ETF is generally healthy and a short term average to indicate a pullback the conditions can be written as follows:

  • Today’s close price is above the 200 day SMA
  • Today’s close price is below the 5 day SMA
  • The high price of 2 days ago is below the high price of 3 days ago
  • The low price of 2 days ago is below the low price of 3 days ago
  • The high price of 1 day ago is below the high price of 2 days ago
  • The low price of 1 day ago is below the low price of 2 days ago
  • The high price of today is below the low price of 1 day ago
  • The low price of today is below the low price of 1 day ago

EdgeRater Chart Script

To run the strategy within EdgeRater we need to create a series selector using the above criteria. In this case I will write some Chart Script and assign the final selection to a variable named ‘Event’ – this will allow me to drag and drop the script onto the Series Selector design surface and run script over all of the stocks in the snapshot. The code for the ChartScript is:


Above200:= C > SMA(C,200);
Below5:= C < SMA(C, 5);
HL3:= (Ref(H, 2) < Ref(H, 3)) & (Ref(L, 2) < Ref(L, 3));
HL2:= (Ref(H, 1) < Ref(H, 2)) & (Ref(L, 1) < Ref(L, 2));
HL1:= (H < Ref(H, 1)) & (L < Ref(L, 1));

Event: Above200 & Below5 & HL3 & HL2 & HL1;

EdgeRater symbol list

In his book, Larry claims to have tested his strategies on 20 popular ETFs as follows:

DIA, Diamonds Trust (ETF)
EEM, iShares MSCI Emerging Markets (ETF)
EFA, iShares MSCI EAFE Index (ETF)
EWH, iShares MSCI Hong Kong Index (ETF)
EWJ, iShares MSCI Japan Index (ETF)
EWT, iShares MSCI Taiwan Index (ETF)
EWZ, iShares MSCI Brazil Index (ETF)
FXI, iShares FTSE/Xinhua China 25 Index (ETF)
GLD, SPDR Gold Trust (ETF)
ILF, iShares S&P Latin 40 Index (ETF)
IWM, iShares Russell 2000 Index (ETF)
IYR, iShares Dow Jones U.S. Real Estate (ETF)
QQQQ, PowerShares QQQ Trust (ETF)
SPY, SPDR S&P500 (ETF)
XHB, SPDR S&P Homebuilders (ETF)
XLB, Materials SPDR (ETF)
XLE, Energy Select SPDR (ETF)
XLF, Financial Select SPDR (ETF)
XLI, Industrial SPDR (ETF)
XLV, Health Care Select SPDR (ETF)

To use this list within EdgeRater, create a text file containing the above symbols and names (copy and paste the above to a new text file) and add it to the {EdgeRater}\Symbols Lists directory in your documents folder.

EdgeRater data snapshot

Once the list has been created you can easily create a snapshot containing all of the data for those symbols. I built my snapshot using the built-in free Yahoo! data provider. I chose a date range from 1/1/1993 – 12/31/2008.

Running the Analysis

Using the Edge Runner tab within EdgeRater, I selected my freshly created snapshot and checked the box ‘use global exclusion lists’ so that the _INDEX of the snapshot was not included in the results.

Results
————————————————————————————————–
Article Update: Find these ETF Signals with my new program
————————————————————————————————–

It was interesting to see the Event Grid containing events for all of the symbols in the snapshot dating back to 1993. On my laptop running Windows 7 this result was generated in 1/10th of a second – not bad for a sub $700 dollar laptop, and excellent for allowing me to experiment with different parameters.

Here’s a screen shot of part of the event grid where you can see the symbols that EdgeRater has selected. I chose to capture a screen shot of the 07/18/2007 column because that is the date that Larry uses for the EEM example ‘event’ in his book:

Event grid after the 3-day high/low selector has run

Event grid after the 3-day high/low selector has run

Shameless plug…

I find this grid extremely helpful because not only does it show me the stocks that met the criteria today, but it also shows the stocks that met the criteria for every historical day in the analysis period. In other scanning applications that pick out stocks that met criteria for today only, I’m often left wondering what is going to happen to those stocks and trying to remember to come back to my computer in a few days to find out. EdgeRater shows me all the results for all previous days in the grid and I can just double click on any of those items to see a stock chart with the event highlighted so that I can see what happened next.

Statistics
Using the Results tab I took a look at the statistics of the selections and found out what the average gain was after ‘N’ bars of the event occurring. Here’s a screen shot showing the statistics of gain/loss 4 days after the event:

3-Day High/Low gain loss frequency

3-Day High/Low gain loss frequency

The mean value was +.5474% with a standard deviation of 2.8719

Baseline

The gain/loss frequency chart looks really interesting, showing an average of .5474% gain in 4 days. However, that gain might have been due to the general market behavior and nothing to do with the chosen entry points and so I ran a baseline scan which shows the average of all stocks, not just the ones that met the criteria. Here’s the result of the baseline scan:

Gain Loss Frequency - Baseline

The mean value was +.0944% with a standard deviation of 3.4776, so the baseline scan does have a positive expectation when holding for 4 days which indicates that the market as a whole was gaining during the analysis period, but that gain is far lower than the gain when taking the 3-day high/low entry.

Summary

The first strategy in Larry Connors’ ‘High Probability ETF Trading book’ – the 3-Day High/Low Method looks like it produces entries with a positive edge compared to a baseline entry point. This is the first test that I would run on any proposed strategy and in this case it warrants further investigation. EdgeRater is a great tool for designing and exploring many different strategies and I would encourage the reader to view the other reports that are available in the application such as the simulated trade reports including the daily equity chart.

I am now interested in running an analysis of the short entry method for the same time period – if that also has positive expectation I may well be running these on a nightly basis to find trade candidates for my trading account.

To have analysis articles like this delivered to your inbox
sign up here

Chris White
CEO, EdgeRater LLC

{ 5 comments }

EdgeRater 3.3 Released

by Chris White on November 29, 2009

I am excited to announce the release of EdgeRater version 3.3. This new release makes using EdgeRater even easier. Here’s a list of the features:

  • Indicators and scans are listed in a library on the left hand side of the screen. The system indicators and scans are in their own group as are your own custom built indicators and scans (“My Scripts”)
  • Indicators and Scans (aka ChartScripts) can now be applied to a chart area via a drap and drop operation.
  • As each chart area is selected via a mouse click, the chart scripts in that area are listed on the left hand side of the screen allowing you to modify and instantly see the resulting chart – no more right-clicking and opening up separate dialogs to do this.
  • The values panel is now available in its own separate window which can be detached and floated anywhere on your system – even on a second monitor, so that it won’t obscure the chart.

I believe these improvements make a tremendous difference to the whole charting experience and encourage you to give it a try.

Also new – on the website (http://www.edgerater.com) there is now a ‘Feedback’ tab which you can use to ask a question, suggest an improvement, report a problem or give praise. So try out the new features and post a comment on the feedback tab – everyone will be able to see your feedback and you will be able to see everyone else’s.

Fire up EdgeRater and you will automatically get the latest version.

Hope you enjoy these improvements,

Regards,
Chris White,
CEO
EdgeRater, LLC

{ 0 comments }

Using iPhone SIM in PDC09 Acer laptop

by Chris White on November 22, 2009

At the Professional Developer Conference 2009 (PDC09), all attendees were given an Acer laptop that had been specially built for Microsoft. It has Windows7, 64 bit dual core proc, 250GB hard drive, wireless, bluetooth and 3G. Now it seemed to me that a good percentage of attendees were also carrying around an iPhone – hard to say how many but if I had to guess I would say at least 30%. So obviously there will be many attempts to insert said iPhone SIM into said laptop in hopes of browsing the inter web.

When the PDC ended I flew into Vegas for the Las Vegas Trader’s Expo, and have since had some time in the hotel room to get this scenario working. The key piece of information seems to be the DNS server IP address settings.  The successful approach I have found is as follows:

Download and install the HUAWEI 3G driver from the Acer website, also useful is the Acer 3G connection manager software which you will find in the same download area.

Insert iPhone SIM into laptop (socket is underneath the battery)

In Windows 7, open up the wireless connection list (where you normally see a list of SSIDs available for connection). You should see a new group called ‘Mobile broadband connection’ with a single entry called ‘AT&T’.

Right-click on the AT&T entry and choose the profile tab. The setting that appears to be important here is the Access Point Name (APN). I have: wap.cingular

For the username and password entries, I found that it doesn’t matter what you use. Web searches will tell you that combinations of WAP@CINGULARGPRS.COM and password CINGULAR1 are good, but i found that also ‘foo’ and ‘bar’ work just as well.

Next, set up the IP settings, go to

Control Panel\Network and Internet\Network Connections

right-click->properties->IPv4->obtain an ip address automatically (no need to set the DNS settings at this point as they appear to be overwritten with 0.0.0.0 when you make the connection)

Now right click the connection and select connect/disconnect, this brings up the popup dialog with the AT&T entry. Click this and click ‘connect’.

Once the connection has been made, go back to the IPv4 properties and manually enter a DNS and alternate DNS IP address (use your favorite one).

Click OK and voila, the internet connection should come to life.

Every time you disconnect/reconnect you need to manually change the DNS settings, which is a pain and hopefully someone will let me know a better way.

Hope this helps.

{ 2 comments }

Trader's Expo Las Vegas 2009

by Chris White on November 20, 2009

Here at the 2009 Las Vegas Trader’s Expo, taking a look around at the old and new products and listening to the hype. The Expo seems smaller this year, and also it is now on the 2nd floor instead of the 3rd. Makes choosing the escalator over the elevator more practical.

This morning I took a quick breeze through the exhibition hall and stopped by to talk to some of the exhibitors. The more I see the more I get excited about EdgeRater. The features and performance and graphics in EdgeRater far outclass anything here including a yet to be released product that is going to be in ‘limited availability’ in December.

I think there is a real advantage to being the sole person responsible for the software development plan and also being a trader and using the software for my own trading.

These other guys with developers who don’t trade, and multiple salesmen who may or may not communicate the feedback from these valuable customer contact events to the development team are at a real disadvantage when it comes to getting the needed features into the product and providing real innovation.

{ 1 comment }

PDC Windows 7 Boot Camp

by Chris White on November 16, 2009

Spent the day at the LA convention center attending the pre PDC workshop ‘Windows 7 Boot camp’. Online this workshop was listed as Sold Out, but the auditorium was really less than half full. Not sure if that is because people signed up for this free workshop and then were’t concerned about dropping out later, or if the presentation was moved to a larger room due the number of responses.

The morning session from 10am to Noon covered some of the under-the-hood features that are not surfaced as whizzy features, but every user will appreciate as they use their computer in a regular way. I’m talking about the performance improvements and the two main areas that were discussed were the kernel scheduler and the memory manager. Whilst I will hopefully never have to fully understand the intricacies of this work, the presentation left me thinking that I am very thankful for it having been done. The two presenters (a Microsoft Principal Architect, and a Microsoft Distinguished Engineer) delivered extremely good presentations which showed why they have achieved the positions they have. The work that they have done is complex and I can only imagine the team effort that was involved in design, implementation and testing. If I was wearing a hat I would take it off to those guys!

Lunch was provided in an enormous hall that was set up to hold all the conference attendees once it gets fully underway tomorrow. I’m guessing there was space for over 2000 lunchers, although for the pre-conference there was probably only 900 or so. Still, it was a vast space and very well organized – let’s see how they do when the crowd arrives tomorrow.

The afternoon sessions dealt with features that regular users can actually touch and feel, and included the ‘Task Bar’, Multi-touch features, DirectX graphics, Sensor aware API, and ended with a presentation of each of the 7 winners of the Code 7 Challenge – each entry had to demonstrate an application where some of the new Windows 7 features were used. My favorite was a game similar to War Craft, but where the armies were controlled by multi-touch gestures and a paint brush. This looked really cool.

During the afternoon I went to get EdgeRater logo certified for Windows 7, and have to wait now to find out the results – apparently a Click Once application such as EdgeRater already passes most of the requirements due to the Click Once deployment technology taking care of installation.

So, day 1 is complete, now must brave the hoards for the next few days before heading to Vegas for the Las Vegas Trader’s Expo.

{ 0 comments }

Insights from John Bollinger

by Chris White on November 15, 2009

I recently watched a video in which John Bollinger, the creater of the Bollinger Bands discussed the creation of the bands and gave some insight on how he uses them. I hadn’t previously heard John Bollinger himself speak and so it was very enlightening.  Here’s the link incase you are interested:

https://tdameritrade.omnovia.com/archives/37667

You’ll have to enter your name and email address on that site in order to see it, but it is worthwhile if you haven’t heard John Bollinger before.

I didn’t realize that Bollinger Bands are 26 years old this year – thought they were older than that.

Of course during the video I couldn’t help myself think of EdgeRater and how some of his charts could be displayed and so I did a bit of arranging and came up with a chart that mimics what John Himself finds useful for his trading:
Also at time code 7:49 in the video he makes reference to how Bollinger Bands can be used for indicator normalization, and the example he gives is how the RSI signals of level 30 and 70 can be improved by using a 50 period Bollinger Band of RSI and then plotting %B of that. I put that script together and the output can be seen here:


My script that produced that chart is:
WRSI:= RSI(C, 14);
MID:=  MA(WRSI, N);
UPPER:=  MID + P*STDP(WRSI, N);
LOWER:=  MID – P*STDP(WRSI, N);
B : (WRSI – LOWER) / (UPPER – LOWER);
Fire up EdgeRater and enter this formula – you’ll be able to see the difference yourself.

{ 1 comment }

Force Index Pullbacks – Is there an Edge?

by Chris White on October 29, 2009

The Opportunity

I was fortunate enough to be asked to present EdgeRater at the semi-annual HGSI seminar in Palos Verdes, LA. The seminar is a 3 day event where Ron Brown and Ian Woodward present market analysis and techniques for using HGSI (an excellent trading tool BTW). On day 2 I had the hour before lunch to demo EdgeRater. I chose to demo a technique where HGSI and EdgeRater can be used together to get a really powerful combination of tools.

Backtesting

The technique I used was to backtest a list of stocks that had been picked out by HGSI back in June. HGSI is great at analyzing current market conditions and ranking stocks based on fundamental and technical characteristics. I used a list that HGSI had generated back in June consisting of ‘leaders’ – stocks that showed great strength both fundamentally and technically. I loaded that list into EdgeRater and created a snapshot of data containing those leaders, with data going back 2 years just to ensure that the technical analysis functions had enough data to start producing values the day after the list was generated in June.

Force Index

Ron had been discussing the applications of Alexander Elder’s Force Index with the attendees the day before my presentation and so I wanted to see if there was a statistical edge in using a Force Index pullback entry on that group of stocks during the period of time after the list was generated to present day.

I constructed a Force Index pullback script for EdgeRater, and then ran that over the list of June’s leaders from June to the day prior to the seminar, picking out every pullback event in the list during that period of time. I then looked at the statistics to see if over a 1 bar, 2 bar, 3 bar … 10 bar period a long entry would have produced better results than just buying the average of all the stocks in the list and holding them for the same number of periods.

Results

The results were quite interesting. It turned out that the only hold period that performed better than buying the average was a 1 bar hold – meaning to purchase the stock on the open the day after the Force Index pullback event was generated, and selling it on the close of that same day. All other hold periods performed worse than average, potentially indicating that there is an edge in shorting those events.

Conclusion

The idea would need more investigation – I only conceived of it the evening prior to my talk after watching Ron’s presentation. My selector was also very basic and detects the condition of 13 period Force Index being positive whilst the 2 period Force Index crossed from negative to positive, indicating a longer term bullish condition and a pullback and resumption of a short term bullish indicator.

This is just one idea that you could take further with EdgeRater analysis, perhaps using a group of stocks from an index rather than being fundamentally generated, and perhaps honing in on the conditions for a better percentage gain overall.

{ 0 comments }

Statistical Edge

by Chris White on April 8, 2009

Stock market anomalies occur every day, and if you knew where to find them you could enjoy outstanding returns year after year.

Question:
Would the following trading idea have been profitable in 2008?:

  • Every day at market open, buy the two S&P 500 stocks that lost the most percentage points the previous day
  • At market close, sell those 2 positions

Answer:

Percentage gain/loss of above scenario

Percentage gain/loss of above scenario

There would have been 504 stock purchases with only 2 positions open at any one time, and if you had bought a fixed dollar amount of $1000 of each stock, by December 31st 2008 you would have gained $628, even though 2008 was a very bad market for stock buyers overall.

If you have a curious mind you are now probably wondering:

  • What if I held the stocks for 2 days, how about 5 or 10?
  • What if instead of buying the worst percentage losers I bought the best percentage gainers?
  • What if I shorted the best percentage gainers?
  • What if I forgot about the very extreme gainers and traded stocks in the upper gain bracket?
  • What if I only traded stocks that met my price and volume criteria?

Answering questions like these will lead you down the path to designing a great trading plan and becoming a profitable trader.

{ 0 comments }

← Previous Entries

Next Entries →